Impact of Intermediate-Task Training on Zero-Shot Cross-Lingual F1 Scores Across Language Families

Intermediate-task training---fine-tuning a pretrained model on an intermediate task before fine-tuning again on the target task---often improves model performance substantially on language understanding tasks in monolingual English settings. We investigate whether English intermediate-task training is still helpful on non-English target tasks. Using nine intermediate language-understanding tasks, we evaluate intermediate-task transfer in a zero-shot cross-lingual setting on the XTREME benchmark. We see large improvements from intermediate training on the BUCC and Tatoeba sentence retrieval tas

Research goal: What is the impact of applying intermediate-task training with varied model sizes (e.g., 7B, 13B, 30B parameters) on zero-shot cross-lingual F1 scores for different language families (e.g., Indo-European, Afro-Asiatic) on the XTREME-R benchmark?

Autonomous synthesis report generated by Assignee Research. Tribunal consensus score: 8.8/10.